Dimensional or Categorical Approach to Tinnitus Severity: an Item Response Mixture Modeling Analysis of Tinnitus Handicap

Abstract

Whether handicap due to tinnitus-sound(s) in the ears and/or in the head in the absence of an external auditory source-is best conceived as dimensional or categorical remains an unanswered empirical question. The objective was to investigate whether tinnitus severity was best conceptualized as qualitatively distinct subtypes, quantitative differences varying along a single continuum, or as severity differences within subtypes. Various forms of item response mixture models (latent class models, factor analysis models, and hybrid models) that corresponded to the competing hypotheses were fitted to item responses on the Tinnitus Handicap Inventory in a Swedish sample of individuals with tinnitus (N = 362). A latent class model could be fitted to the data with a high probability of correctly classifying individuals into three different classes: high-, moderate-, and low-severity classes. However, a comparison of models showed that a unidimensional factor analysis model with a single class provided the best fit to the data. The analysis provided evidence that tinnitus severity varies along a single severity continuum from mild to moderate to severe tinnitus-related handicap. The result that tinnitus severity exists on a continuum rather than as discrete categories has important implications for clinical research.